Re: [RFR] Description for package quassel

[Jonathan Nieder <jrnieder@gmail.com>]

Hi,

Justin B Rye wrote:
> Thomas Müller wrote:

> Maybe...
> 
>    Quassel is a modern, cross-platform, graphical IRC client made up of a
>    "core" component, which maintains a connection to the IRC server, and
>    one or more "clients", which can attach to and detach from a local
>    or remote core. This gives some of the same advantages as using screen
>    with a text-based IRC client.

For what it’s worth, I like this.  I would leave out the
“cross-platform” (what is more relevant is that it is graphical, and
most software packaged for Debian is cross-platform these days).
Though probably it is worth mentioning that it works well on a cell
phone.

The use may still wonder “why?”, so I think you should mention the use
for logs.

 When no clients are connected to the core, it stores messages in a
 database so they can be read later.

> Maybe with the word "channels" in there somewhere.  And surely
> there's a label somewhere that describes this design... modular?
> Master/slave?  Or perhaps it's an IRC multiplexer...

Maybe:

 graphical multiplexing IRC client - core
 graphical multiplexing IRC client - Qt front-end
 ...

>> Well thats for all the people out there - like me;-) - who have no interest in 
>> viewing the backlog. I just want an irc client - I use the monolithic client.
>
> So what advantage does it have over any other graphical IRC client?

It looks like the layout is roughly based on WeeChat’s, so it might be
comfortable for WeeChat veterans that want something graphical.

Comparing <http://quassel-irc.org/node/104> with the screenshots from
<http://lwn.net/Articles/216456/>, one benefit seems to be that it
displays multiple channels at once in a single window, rather than
using the insane tabbed layout some others use.  But that is only my
quick and uninformed impression.

> Remember that the main job of a package description is to let admins
> know whether they want to install a given package.  It seems odd
> that the package that holds the unmodified name "quassel" is the
> version with Quassel's key feature stripped out...

I agree: wouldn’t a name like quassel-monolithic do the trick better?

Just my two cents,
Jonathan